The present disclosure relates to networking and, more particularly, to improved network utilization in policy-based networks where such networks may be dynamic.
Policy-routed networks with multiple paths between a source and a destination are generally not optimized to take advantage of those multiple paths, artificially reducing the throughput of the network, particularly in situations where topology and link quality vary over time. If a link is saturated for a given quality of service (QoS) level or differentiated services code point (DSCP) value, additional traffic is dropped even if a suitable alternate path is available. In dropping data, critical information may be lost or time-sensitive information may be unduly delayed. This is especially important in tactical networks where loss or delay may have extreme consequences.
Conventional network managers have rules for policy-based management that determine the routes that should be taken based on QoS level or DSCP value, but if those routes are congested, data is typically dropped.